Predicting infrastructure damage and economic impact of sinkholes requires high accuracy mapping distribution and development. The study mapped sinkholes and sinkhole hotpsots in Johnson City, TN using LiDAR-derived Digital Elevation Model (DEM) and a database of known sinkholes which were matched to LiDAR-derived depressions. For all matched depressions (n = 404), three metrics were calculated: circularity index, ratio of length to width of the Minimum Bounding Rectangle (MBR) and percent coverage of the MBR by the depression, and 3,634 new sinkholes were identified. Newly developed hotspots were identified in north Johnson City and other areas in the south near the Johnson City Medical Center. The methodology developed can be applied to identify hotspots in other small metropolitan cities and the hotspot map produced can be employed in hazard mitigation planning, resource allocation, and made available publicly to property owners and insurance companies.

Improving Sinkhole Mapping Using LiDAR Data and Assessing Road Infrastructure at Risk in Johnson City, TN. Kingsley Fasesin1, Dr. Ingrid Luffman 1, Dr. Eileen Ernenwein 1 and Dr. Arpita Nandi1 1 Department of Geosciences, College of Arts and Sciences, East Tennessee State University, Johnson City, TN; Abstract Predicting infrastructure damage and economic impact of sinkholes along roadways requires mapping of sinkhole distribution and development of a model to predict future occurrences with high accuracy. The study is carried out to define the distribution of sinkholes in Johnson City, TN and risks they pose to roads in the city. The study made use of a 2.5 ft Digital Elevation Model (DEM) derived from Light Detection and Ranging (LiDAR) data acquired from Tennessee Geospatial clearing house (TNGIS) and an inventory of known sinkholes identified from topographic maps. Depressions were identified using the LiDAR-derived DEM by subtracting a filled-depressions DEM from the original study area DEM. Using a spatial join, mapped sinkholes were matched to depression polygons identified from the LiDAR-derived DEM. For all matched sinkhole-polygon pairs, three indices were calculated: circularity index, area ratio of minimum bounding rectangle, and proximity to train tracks and roads. The dataset was partitioned into training (70%) and validation (30%) subsets, and using the training dataset, thresholds for each index were selected using typical values for known sinkholes. These rules were calibrated using the 30% validation subset, and applied as filters to the remaining unmatched depression polygons to identify likely sinkholes. A portion of these suspected sinkholes were field checked. The future direction of this research is to generate a sinkhole formation model for the study area by examining the relationship between the mapped sinkhole distribution, and previously identified sinkhole formation risk factors. These factors include: proximity to fault lines, groundwater and streams; depth to bedrock; and soil and land cover type. Spatial Logistic Regression analysis will be used for model development, and results will be used to generate a sinkhole susceptibility map which will be overlain on the road network to identify the portions of interstate and state highways at risk of sinkhole destruction.